A Novel Data Segmentation Based Approach for Meter Topology Identification using Smart Meter Voltage and Power Measurements

Authors:
  • HLHan Pyo Lee
    Han Pyo Lee
  • PRPj Rehm
    Pj Rehm
  • MMMatthew Makdad
    Matthew Makdad
  • EMEdmond Miller
    Edmond Miller
  • NLNing Lu
    Ning Lu

Abstract
This paper presents a data segmentation based approach to improve identification accuracy and reduce false-positives in meter topology identification problems, such as meter phase identification and transformer-meter pairing identification. Using smart meter data as inputs, voltage correlations between meters are calculated to determine which phase a meter is on and which transformer it is supplied by. The underlying assumption is that voltage profiles of meters supplied by the same distribution transformer or on the same phase are strongly correlated. In this paper, we first derive that i) when two customers consume at different power levels, the voltage drops on the secondary circuits that connect the customers to the transformer can be significant, and ii) large secondary circuit voltage drops can significantly weaken the voltage correlation between meters on the same phase or supplied by the same transformer. Based on this discovery, we proposed a data segmentation based approach that uses power bands for highly correlated voltage segment selection in order to minimize correlation deterioration. Thus, instead of using the entire voltage waveforms, we use an ensemble of voltage segments for voltage correlation calculation. The proposed methods are developed and tested on both synthetic and real feeder data sets for two algorithms: meter phase and transformer-meter pairing identification. Simulation results show that the proposed algorithm outperforms the state-of-the-art methods in both accuracy and robustness.